TL;DR

An updated conversion formula has been implemented to better map cgroup v1 CPU shares to cgroup v2 CPU weight, improving priority alignment with non-Kubernetes processes and offering finer granularity for workload management.

What happened

['Kubernetes introduced an enhanced conversion formula from cgroup v1 CPU shares to cgroup v2 CPU weight', 'The new formula addresses issues related to priority allocation between Kubernetes workloads and system processes']

Why it matters for ops

['Improves resource prioritization in multi-process environments', 'Allows for more granular management of workload resources within containers']

Mitigation

  • Upgrade to supported versions of runc and crun that implement the new formula
  • Test changes in a non-production environment before full-scale deployment

Action items

  • Review resource management strategies within applications
  • Update custom tools and monitoring systems for compatibility with the new conversion formula

Detection IOCs

  • Unexpected CPU weight values after upgrading container runtime
  • Decreased performance or priority issues with Kubernetes workloads

Source link

https://kubernetes.io/blog/2026/01/30/new-cgroup-v1-to-v2-cpu-conversion-formula/